Table 2.

Comparisons among different QSAR models on FRAP dataset ^a.

Descriptors	Before Logarithmic Transformation			After Logarithmic Transformation
	Q2	R2	optPC	Q2	R2	optPC
VHSE	0.0042	0.2655	3	0.4878	0.6122	6
5Z-scale	0.1408	0.3177	2	0.4809	0.5568	3
DPPS	0.0059	0.2290	3	0.4147	0.5463	4
ST-scale	0.0263	0.3220	8	0.3968	0.5410	9
FASGAI	0.0470	0.2753	2	0.3735	0.5006	4
E-scale	0.0560	0.2521	1	0.3714	0.4734	5
HESH	0.0444	0.2818	10	0.3668	0.5290	3
HSEHPCSV	0.0259	0.2475	7	0.3624	0.4952	3
G-scale	0.1066	0.2334	5	0.2836	0.3850	1
VSW	0.0130	0.3071	1	0.2382	0.4361	2
MS-WHTM2	0.0342	0.0370	3	0.1728	0.2594	3
MS-WHTM1	0.0452	0.0329	9	0.1207	0.1941	4
T-scale	0.0682	0.0706	2	0.0750	0.2129	10
V-scale	0.0293	0.0748	4	0.0699	0.1495	1
Z-scale	0.0052	0.1445	1	0.0301	0.1456	6
ISA-ECI	0.0242	0.0141	1	0.0071	0.0411	1
Integrated descriptors	0.1069	0.4212	3	0.4953	0.6423	3
BOSS				0.6088 ± 0.0041	0.6655 ± 0.0094	3.5100 ± 2.5086

^a R² is the coefficient of determination; Q² is the cross-validated R²; optPC is optimal principal components for PLS regression model; the results of BOSS are shown in the form of mean value ± standard deviation in 100 runs, the top ranked Q² scores were marked in bold.